Allow BatchNorm on CUDA with track_stats=False #2427
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
2 tasks
|
The "buildkite/flux-dot-jl" tests appear to be failing because NNlib 0.9.13 was installed. 0.9.14 includes my related fixes to NNlib for track_stats=False. Is there a preferred way forward here? Should I increase the compat version of NNlib from 0.9.1 to 0.9.14? |
|
Yes, please do and then this should be ready if tests pass. |
Enables BatchNorm without track_stats in training and test modes. Also, unit tests are added to ensure the CUDA implementation matches the CPU implementation. Also, update required NNlib to 0.9.14 which includes fixes to batcnorm when track_stats=False.
paulnovo
force-pushed
the
batchnorm-no-track-stats
branch
from
7bef09b to
7284b07
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Enables BatchNorm without track_stats in training and test modes. Also, unit tests are added to ensure the CUDA implementation matches the CPU implementation.
This address #1606 but requires changes in NNlib PR 576.
PR Checklist